To prevent shortage of storage space in a service system, an administrator usually set per-user quota as an upper limit of usable space for each user. To avoid service failure caused by resource exhaustion, the administrator tends to set a conservative quota value such as the storage capacity divided by the expected maximum number of users. In this research, we analyzed long-term storage usage history of our email system and file sharing system in Kyushu University. Mostly through the analyzed period, the usage pattern showed a long-tailed distribution similar to log-normal distribution. Also the overall storage consumption slowly increased during the analyzed period. Based on these analysis, we defined "storage utilization ratio" to evaluate how the storage was effectively used. By approximating a storage utilization pattern as a power-law distribution, we proposed a method to calculate the optimal quota value to maximize the utilization ratio.
To prevent resource (especially storage) shortage, information systems such as storage services and email services usually impose an upper bound of resource consumption (quota) per user. In a conservative way, an administrator tends to set a quota value such as the storage capacity divided by the expected maximum number of users for safety and fairness, but it tends to leave large unused storage space, because the users' storage usage pattern shows a long-tailed distribution. In this paper, we analyzed storage usage distribution of some email services to approximate the distribution using a power-law distribution, and proposed a method to calculate an optimal quota value from a target size of storage consumption to increase storage utilization. We applied an optimal quota value we calculated to a real email service and analyzed the effect of quota change. Then, we analyzed actual distributions further to find a better model to approximate the distribution, and found that a lognormal distribution explained the distribution better than powerlaw. We also analyzed two other universities' email service to find similar distribution in these systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.